KWin On Mir: A Solution To Non-Existent Problem

[aseigo]

the question

aha, so why they started wayland then?

There is a need for something after x11. Everyone agrees on this. Wayland has been worked on across the broader community to be that thing. So the question no longer exists, and the reasons given for Mir rather than Wayland don't hold water.

Ergo the statement.

[AJSB]

Purely personal opinion, but I stopped minimizing and maximizing windows ever since OS X's Expose was cloned in Compiz, KWin, Gnome 3 (for Linux) and third-party tools (for Windows) and I find that I'm much more productive than before.

YMMV, of course.

Well, indeed MMMV, for *me* and the way i'm used to work, minimize/maximize is mandatory...

[AJSB]

Who has said that it is usable for everyday use? I have read every post on every thread on the subject and don't recall seeing anyone say that. I might have missed it in a post or two, but it is hardly a common sentiment here. They said it has had a stable release, has a stable API, and actually has something running, none of which Mir has. But nobody has said that everything it needs is present.

It has many, if not most, of the features that are needed, but certainly not all. If it did, it would be done, which nobody has claimed. This is in contrast to Mir, which has essentially none of the features.

That being said, phones don't need this, so it probably won't affect Tizen.

OK, English is not my natural language so i might got the wrong idea, but, IIRC...one of the devs (or even main dev) of Wayland said that Wayland was perfectly usable or something like that, if we use XFCE (witch i do) but if minimize/maximize , a very basic feature not only to me but for majority of users (my assumption) in special anyone coming from M$ OSs, is far from working properly then Wayland is not usable at all , not even for XFCE.

[Ericg]

OK, English is not my natural language so i might got the wrong idea, but, IIRC...one of the devs (or even main dev) of Wayland said that Wayland was perfectly usable or something like that, if we use XFCE (witch i do) but if minimize/maximize , a very basic feature not only to me but for majority of users (my assumption) in special anyone coming from M$ OSs, is far from working properly then Wayland is not usable at all , not even for XFCE.

I actually just tried it today on Arch

Things that dont work:
~Minimize supprot, obviously (maximize/unmaximize does though)
~X apps-- because XWayland hasnt been merged, unfortunately
~Native environments since they require porting. (I didnt have a qt5 app available to test Lighthouse support)
~Touchpad is sloooooooooow (Acceleration is currently disabled, Daniel Stone said he was working on it, or atleast he was supposed to be working on it)

You get a desktop, you get a terminal, you get a date and time notification, you get a few examples, 2 working openGL examples (glxgears and a seperate opengl triangle) vsync works, There's a pdf reader, an image viewer, terminal as i said. Its very barebones. Ya need gtk3 apps (not sure what version of gtk3 as proper wayland support) and qt5 apps to really use it. But its THERE, and its WORKING.

[Xeno]

I'd be really surprised if Mir would succeed.
In order for it to happen Canonical needs to
a) write display server that fulfills demand of all major DEs
b) would have ability to run legacy apps
c) would be supported by all major toolkits.
d) would have several major advantages over Wayland.

Then Canonical would have to convince developers of ail major DEs to switch to Mir.
So in the end other distros would use it.

So lets analyze those conditions.
a is probably possible but it'll take lots of work (and time). B might kinda work (nested x) but here's little quirk that we cover in point c. C - we know it's not gonna happen, Canonical expressed interest in Qt 5 and Gtk3 with Qt3/Qt4/Gtk2 left out. D. is extremely unlikely - there might be some minor things (if any, but nothing major, than can't be done in Wayland).

So let's look at possible adoption by DE's
- Unity - will support for sure
- KDE - we know their stance already (not before adoption by most of the distros)
- GnomeShell/Gnome3 - unlikely (esp. looking how it's influenced by RH)
- Cinnamon - maybe but I really doubt as it's Gnome3 derivative so it most likely would follow GS and most likely it's too much hassle for little gain.
- Mate/Xfce/Lxde - known to be not working (no support for Gtk2)

Now let's look at which distro would hop into Canonical bandwagon.
- Fedora/RHEL/CentOS/ScietificLinux/Unbreakable Linux - they would need to convince RedHat. I can't really see it happen. RH is known to show NIH syndrome too, so if there are two similar solutions RH will always pick one they can influence. Also I believe RH already has invested some developers work int Wayland, what makes switch even less likely.
- Suse/Debian/Gentoo/Mandriva/Mageia/Arch - none of them use Unity as their main desktop, having their own ports of GS and trading legacy DEs is rather too much costs for supporting Mir. That also means no support for Mir in upstream KDE, so it's another point against Mir.
- Mobile distos (Tinzen/Mer) - Intel already said they go Wayland, so no place for Mir here.

Regarding the proprietary drivers vendors, they'll go where RH goes. Ubuntu might be strong on average desktop,but AMD/NVidia interests are enterprise systems where so far RH is largest player.

BTW How the hell Canonical wants to support apps like Gimp/Inkscape/LibreOffice - they all Gtk and I really doubt Canonical will port'em all to Gtk3 on their own.
Mir smells like complete disaster to me.

[edit]
On second thought, my guess is legacy apps will be run in nested X, but it somehow makes Mir redundant when your main productivity apps still need X


In the end IMHO all this Mir hype is badly damaging Linux as a whole., Display server is critical component, that need to bo common for all distros. the reason X is there is not that X server is so great, but it's there because display servers are extremely important and huge PITA to replace.

[Hermit]

From what I understand, the amount of work still required to make Mir or Wayland compatible with Android drivers is... exactly the same. That's because they have the same basic architecture and are both EGL based. As said many time, Wayland has been ported to Android last year as a proof of concept, to show that even if they didn't aimed for it, it would still be easy.

As such, the people complaining are not missing much, and you're spreading BS.

Exactly right, no where did Mir claim that it's binary compatible (or planning to be binary compatible) with existing Android graphics driver. It only states that it's compatible with the Android graphics driver *model*, e.g. it's EGL based. Well, guess what, so is Wayland! So the notion that Mir is more complete or closer to production use on mobile devices than Wayland just doesn't hold water. If the goal is to bring a product to market as quickly as possible then you might as well straight up use SurfaceFlinger or fork it. But then it really doesn't fit into the "convergence" story now, does it?

For those who are impatient about Wayland. Think about this: for a truly usable graphics stack, you need more than just the display server. A lot of foundational pieces need to be in place. You need the core protocol stabilization (1.0 since last October), the kernel pieces (KMS/DRM, not strictly required, but pretty important for good performance), Mesa EGL, toolkit support (Qt, Gtk+, EFL, SDL, etc.), and finally the shell protocol. The last piece is the only one that is still in a state of flux. I feel that Wayland may still need a couple more years before it's ready for production use, but all of the foundational pieces are pretty solid.

[AJSB]

So, in that case X will reign for *at least* 2 more years in Linux ?

[GreatEmerald]

So, in that case X will reign for *at least* 2 more years in Linux ?

Hah, that made me remember how crazy the X Window System version release history was, actually. I plotted a chart to show that:

It's pretty much a logarithmic curve, with the limit at +infinity being X12 or so. Quite fascinating.

[AJSB]

Nah....you are incorrect....they simply decided to keep it at X11 because it's a much cooler name than X12, X13,etc.

[GreatEmerald]

Nah....you are incorrect....they simply decided to keep it at X11 because it's a much cooler name than X12, X13,etc.

Nope, not really: http://www.x.org/wiki/Development/X12